@MarcelKrüger @DavidCarlisle @JosephWright why does this error but compiles if one remove the block code?

\RequirePackage{tagpdf-base}
\RequirePackage{latex-lab-testphase-block}
\directlua{function blub () return"\\test" end}
\documentclass{article}
\begin{document}
abc
\end{document}

@UlrikeFischer \show\\

@JosephWright yes during lunch I realised that this in a directly...

@UlrikeFischer :)

In other news, I will match yesterday's dev release of 2e with an l3 one

Although it's not massively exciting

@JosephWright and the question is now, is the kanbun package wrong because it uses \\newif in a directlua or is the block code wrong which makes \\ unprotected?

@UlrikeFischer Oh dear

@UlrikeFischer Well probably both, although making stuff \protected is still relatively new for LaTeX2e

@JosephWright the correct escape is \string\\newif?

@UlrikeFischer Well sort-of - I'd say it's \directlua{require("kanbun.lua")}

@JosephWright there is a require, but for some reason some functions are defined in the sty.

@UlrikeFischer All v odd

@UlrikeFischer But we should have consistent status for \\ in any case

@JosephWright actually I wonder if \string\\newif is correct. Wouldn't that expand the \newif directly? don't you need \string\\string\newif to get the same as with an protected \\ ?

@UlrikeFischer They want (I think) the conditional to be evaluated

@DavidCarlisle Maybe something for you? reddit.com/r/LaTeX/comments/1etjd25/…

(thought so, after reading github.com/latex3/latex2e/issues/432)

@JosephWright \protected :-)

@DavidCarlisle Indeed

@mickep there's an issue about that somewhere, but basically it's your fault

@UlrikeFischer the latter

@DavidCarlisle ;-) so someone who understands the template assignments should correct that ...

@UlrikeFischer can't you just redefine \@normalcr and leave the \protected\def\\ unchanged ?

@DavidCarlisle Tss

@mickep well the user has switched fonts, switched to a totally different math layout system that does not have the specific font parameters for delimiters on generalised fractions, and then expects identical appearance. Do you have a better plan than blaming you?

@DavidCarlisle I thought luatex just used the font parameters. But I do not see where I come in. :)

Expecting the same as pdftex I cannot understand, but the parentheses are bigger than one would expect.

@mickep yes they are also bigger than with xelatex

@DavidCarlisle I don't know why Frank used cr-cmd = \\ in the template code.

@DavidCarlisle Sounds good to me

@DavidCarlisle Like we said, drip-drip of \protected

@DavidCarlisle Yes, but I have no clue what XeTeX does with the parameters.

@mickep opentype math fonts don't have the font parameters corresponding to generalised fractions (luatex fakes them i think) so initially \binom didn't work at all in xelatex but then we made a new definition using normal \left\right stretch delimiters but tuning the box layout to get tehsame layout with latin modern at aleast, but something is missing again in luatex I think

(I have a feeling that it somehow tries to go traditional TeX, but how that conversion goes from opentype parameters to tex parameters, I have no idea about, or even if that is the case)

@DavidCarlisle Oh, now I must confess that I don't know what generalised fractions is really doing...

But you mean that the fraction (parameters) in opentype are more limited than the ones in generalised fractions?

@mickep it was just the primitive \atopwithdelims but that doesn't work at all in xelatex using opentype math fonts so amsmath provides a definition using \left( ... \atop ... \right) with custom local settings to hide the difference I'd need to check why it has luatex specific code these days

@mickep yes for reasons I never really understood, in classic tex, atopwithdelims doesn't use the same stretch parameters as left..right it has custom ones and opentype doesn't really hav ethat it only has one notion of stretch brackets

@mickep github.com/latex3/latex2e/issues/432

@mickep do I really want to sign up to reddit to comment :(

@DavidCarlisle That is the same one I linked to above.

@DavidCarlisle You could ask @UlrikeFischer to comment. But I can also write something (but that will be a blame of you of course)

@mickep so it is, sorry. I haven't checked if the suggestions there fix the case in the reddit post but they probably do, fairly reliable source on that issue.

@DavidCarlisle Maybe someone who has LaTeX and the fonts installed could test. :P

(I did not read the whole issue through, but I thought it was considered to be fixed?)

@mickep add it to my stack of weekend jobs.... maybe if I fix it I'll sign in to reddit and post there

@mickep it's still open and has no fixed-in-dev label so I suspect not.

@DavidCarlisle Ohh, I did not mean to spoil your weekend :(

@DavidCarlisle Ah, I did not read carefully...

@mickep I need to be editing amsmath anyway (to fix \dots that's broken in the github develop branch at present)

@DavidCarlisle Oh no, that sounds worse...

@mickep only if you follow \dots with a primitive tex \if... If you \futurelet\nexttoken\foo and then have a lot of conditional code testing \nexttoken it's interesting in how many ways it can go wrong if \nexttoken ends up being \let to \iffalse

@DavidCarlisle I never understood the point of \dots, seems easier to use \ldots/\cdots/whateverdots directly. And apparently, more safe.

(But of course, you do not use \dots...)

@mickep I suspect it's because the AMS had idiosyncratic rules for which to use and got tired of having to manually edit every author supplied manuscript so they tried to automate it.

@mickep I don't use tex

@DavidCarlisle I assumed so... :)

@mickep don't let @barbarabeeton see this discussion

@DavidCarlisle Dodos are very good at secrets, as @PauloCereda knows.

@DavidCarlisle -- I can't claim to understand that either, but I think the absence of suitable values in opentype can be blamed on Microsoft, which didn't include them in their font implementation because "they didn't need them". I can try to find where this is stated (probably in documentation of Cambria, or ask Murray).

@DavidCarlisle -- too late.

@barbarabeeton Interesting. I'm also curious about what parameters, if any, you are missing.

@mickep amsmath doc says

%    the |withdelims| primitives do not work in xetex with OpenType
%    fonts, and the relevant font dimen parameters are often not set
%    in luatex as there are no matching values in the OpenType Math
%    table, so here we use variants that use the font parameters if
%    they are set, but scale using |\left\right| rather than the
%    |withdelims| primitives.
%

@DavidCarlisle -- \dots is documented in Joy, so it was Mike Spivak, not AMS.

@mickep fontdimen parameters 20 and 21 don't have corresponding values in opentype

@barbarabeeton well true but that's amstex isn't it?

@mickep but I think I only had to modify things for xetex, I think luatex just got changed at the same time for consistency, but now seems broken

@DavidCarlisle -- amstex was first undertaken by Bob Morris; Mike Spivak took it over and polished it, So I can't say for sure whether Bob originated it. That's quite possible. In any case, the AMS editorial complex had no control over it. The only thing they really complained about was the "Spivak pronouns" in Joy.

@DavidCarlisle Is it the \sigma_{20} and \sigma_{21} in Ulrik's article (ntg.nl/maps/38/03.pdf)? If so, 21 seems to be DelimitedSubFormulaMinHeight. (Too lazy to walk 4 meters to look in the TeX book...)

@mickep hmm interesting, more weekend work, i wonder if it was just a xetex bug (I'm pretty sure xetex leaves (or did leave) both as 0 in its default loading of opentytpe math fonts) one reason ams handed amsmath support back to us was to fix things up again to work with unicode tex (several edge case things got broken)

@DavidCarlisle I don't know what is broken and where. Another nice read for you this weekend might be tug.org/TUGboat/tb27-1/tb86jackowski.pdf then. :)

It would almost be easier for you to come to Lutten and talk with Taco and Hans. :P

@mickep apart from generalised fractions, some of the introspection of mathchar just failed with low level errors if anything had been defined with \Umathchardef so just had to extend a few things to cope

@mickep also added \Umathstack... in a few places where it made sense

@DavidCarlisle @egreg is here, eager to join you in a zoom session this weekend.

@DavidCarlisle ooh secrets

Today we ask about white lines.

@mickep whites line are so boring! They would be much more interesting in cyan or pink :)

@samcarter indeed!

@samcarter I always knew you were in favour of colortbl doc.

@UlrikeFischer we can blame you for untagged chess I assume?

@DavidCarlisle hard to resist something so beautiful :)

@samcarter the package and its author

@DavidCarlisle ;-). Can you reproduce the cline error? I can't.

@UlrikeFischer yes

@UlrikeFischer ooh but I have a mix of some local dev installations, let me clean up

$ lualatex-dev cl1
This is LuaHBTeX, Version 1.18.0 (TeX Live 2024)
 restricted system commands enabled.
(./cl1.tex
LaTeX2e <2024-11-01> patch level 1
L3 programming layer <2024-07-20>
 (/home/davidc/texmf/tex/latex-dev/base/article.cls
Document Class: article 2024/06/29 v1.4n Standard LaTeX document class
(/home/davidc/texmf/tex/latex-dev/base/size10.clo))
(/home/davidc/texmf/tex/latex-dev/tools/array.sty)
(/usr/local/texlive/2024/texmf-dist/tex/latex-dev/l3backend/l3backend-luatex.de
f) (./cl1.aux) (/home/davidc/texmf/tex/latex-dev/base/ts1cmr.fd)

@UlrikeFischer ^^

@UlrikeFischer works now

@UlrikeFischer oh no that hasn't picked up latest dev, updating again from fau.de

all OK again